Cocaine abuse has reached epidemic proportions in the United States. Despite the scale of the problem, little is known about effectiveness of treatments. This application proposes to determine the comparative efficacy of two prevalent forms of treatment, inpatient and outpatient, for persons with dependence on cocaine and also to determine relative cost-effectiveness (CE) and cost-benefits (CB) of these two treatments. Other goals are to test hypotheses regarding treatment outcomes derived from social learning and social support theories and to compile descriptive data of the sample under study. The sample will consist of persons who are in the work force (blue-collar to professionals) who seek treatment in the private non-profit sector. The research design , regression discontinuity, is a powerful quasi-experimental alternative to randomized assignment to treatments. Using this design, subjects with severe cocaine dependency and related psychosocial problems will be assigned to inpatient treatment and subjects with less severe dependency will be assigned to outpatient treatment. Subjects will be assessed at treatment entry, in-treatment, and for 12 months following exit from treatment. Two major categories of outcome will be assessed: drug use, including use of cocaine, alcohol and other drugs; and psychosocial functioning, including functioning in the areas of employment, family and other social relations, and health care. Approximately 600 men and women throughout the state of California, from 21-58 years of age, who enroll in inpatient or outpatient treatment for cocaine dependency will be recruited into the study. The study will be conducted at member programs of CALNET, an extensive network of hospital-based chemical dependency recovery services that provides care to residents of over 50% of the counties in California. To allow for comprehensive evaluation of the treatments to be studied, this research will apply economic CE and CB methods to the treatment of cocaine dependency. These economic appraisal analyses will provide critical information for policy and treatment decisions.